This dissertation develops a framework for vehicle automation that combines individual driving preferences with trust calibration. It uses naturalistic driving demonstrations and inverse reinforcement learning to customize automation, models individual trust dynamics, and integrates these components through a human-in-the-loop approach.